Disk storage can be presented over a transmission control protocol (TCP) network for shared usage by a number of different users (clients). Access to this shared storage resource can be made available using appropriate network protocols, such as CIFS (Common Internet File System), NFS (Network File System), or any other similar network protocol such as Data Domain® Boost (DD Boost) which is provided by EMC® Corporation. For clients to access such storage, they must have direct connectivity to the network storage devices. Direct client communication is a preferred method, but is not always feasible if the storage is isolated from the clients, which is often the case due to isolation concerns arising out of security issues or physical requirements. Certain backup applications, such as EMC NetWorker® provide a mechanism for clients to access the shared storage through a central aggregation point, referred to as a storage node. In such systems, remote clients connect to the storage node that reads and writes in order, data to the shared storage on their behalf. This type of system, however, creates bottlenecks since clients must are forced to communicate with a specific pre-determined single storage node to read/write data to the target storage device. Moreover, since read/write operations are handled sequentially, concurrent operations are limited or impractical, as are parallel restoration operations.
Under present systems, clients either require direct access to a target storage device, or must go through a pre-defined single storage node to access the target storage. Multiple clients cannot access the same data unless they use the same storage node. Concurrent access to the same data volume is possible only with direct client communication to the target storage, which by itself may not be possible or desired in a number of cases.